Step Matrix Guidance

[invictus2010]

Love this library! I was using the step matrix method like so:

df.rete.step_matrix(max_steps=24, targets=['Confirmation'])

However, I was looking over a 48 hr period. Does this method have any session logic? For instance if User X went Home -> Product, exited the application, and then returned 4 hrs later to the home screen, this would be treated as his third screen correct? It would not start his journey over--correct?

[tokedo]

You can plot step_matrix by sessions as well!

In this case you need to pre-process your data by creating a separate column in your dataframe with session_ids. You can use function data = data.rete.split_sessions(thresh=14400, session_col='user_sessions') where 14400 is 4hrs in seconds. Function rete.split_sessions() will be more documented in the future updates as well as included in a separate upcoming guide about data pre-processing for Rete in general.

So the solution for you will look like:

df = df.rete.split_sessions(thresh=14400, session_col='user_sessions')

retentioneering.config.update({'user_col': 'user_sessions'})

df.rete.step_matrix();

Again more convenient functions and guides to pre-process data will come soon!

[invictus2010]

@tokedo man you guys have thought of everything!

[boum2078]

Hi @tokedo and all, was the split by session documented yet...it would be interesting to extend this library such that it ll give you average stats on several sessions over an extended period of time !

[ChernyshovAnton]

The commentary answered the original question.

At the end of April 2023, we will be moving to the version 3 of the library. You can use pip to install packages into your beta version: pip install retentioneering --pre

Documentation on the SplitSessions data processor is available for version 3: User guide API Reference